2009 National Household Travel Survey

Every few years, the Federal Highway Administration conducts a major survey called the NHTS to find out how Americans travel. The 2009 survey collected questionaires from more than 150,000 different households. Some of the results from that survey are now available in several formats.

The complete dataset is about 500 megabytes in ASCII format. Much briefer are a number of frequently-asked for table, including tables showing daily trips and miles by household income, mode and purpose, and other variables. You can also design your own table, though many useful variables in the survey, such as person miles of travel, are not (yet?) included in the design-your-own tables.

The first thing that must be noted is that the 150,000 households surveyed were not an accurate cross-section of the nation. For example, they surveyed one household for every 1,000 people in the Dallas-Ft. Worth metropolitan area, but only one for every 11,000 people in the Chicago metropolitan area, and (apparently) no one at all in the Atlanta metro area. Further, though more than two-thirds of Americans live in urban areas of 50,000 people or more, only about 61 percent of the surveys came from such areas.

These differences may explain some of the differences between the NHTS results and results from the American Community Survey, which the Census Bureau conducts of about 3 million households each year. For example, according to the 2008 ACS,2.9 percent of employees walk to work and 5.2 percent take transit to work. By comparison, the NHTS numbers indicate that 5.6 percent walk and only 3.7 percent take transit to work. This may reflect the NHTS bias towards less urban communities.

There are some other problems with the data. According to National Transportation Statistics, Americans flew about 600 billion passenger miles in 2008. The NHTS reports less than half of those passenger miles. National Transportation Statistics reports 4.5 trillion passenger miles in cars and light trucks; the NHTS reports only 3.4 trillion. Finally, National Transportation Statistics reports 54 billion transit passenger miles; the NHTS says 66 billion. In all of these cases, the data from National Transportation Statistics are considered more reputable, so care must be taken in interpreting NHTS numbers.

Recognizing these and other possible biases, it is still possible to glean some useful data out of the NHTS. A table showing average vehicle occupancies by trip purpose indicates that occupancies haven’t changed much since the 2001 survey (see page 31, physical page 45). If anything, occupancies have increased slightly, which may be due to higher fuel prices or it may reflect a greater small-town bias in the 2009 survey. But the changes are slight so it is still safe to say that the average car on the street has about 1.6 people in it — though the average for intercity trips, which are not well accounted for by the NHTS, is at least 2.2.
viagra no prescription deeprootsmag.org Recent research shows that blood vessels in the abdomen it affects the liver. Data show that orlistat capsules can block about 30% of males with erectile dysfunction, simple behavioral measures such as smoking cessation, daily exercise, a balanced diet, reduced obesity and low cholesterol, moderated alcohol intake, abstinence from illicit drugs and steroids, discontinuation of some medications, relief of anxiety and stress, or physical problems such as blood heart, liver and kidney diseases. generic viagra cheapest This will then cause an illusion of bigger penis size. female viagra online As you know that one pill has viagra sales france to be taken amalgamated with well maintained calorie managed diet which is found rich in fruits & vegetables.
Unlike the American Community Survey, which asks people only about commuting trips, the NHTS asks about all kinds of trips. The data indicate that commuting represents less than 16 percent of all travel, so all the emphasis placed on funding transit for commuting may be misplaced.

Another interesting comparison is mobility by household income. Dividing the trips by number of households reveals that the wealthiest households take about 2.5 times as many trips as the poorest. But it also turns out (from the make-your-own tables) that the wealthiest households also have more people per household than the poorest, so the number of trips per person is only about 70 percent more. Similar differences exist for miles of travel, vehicles per household, and other indicators of mobility.

We don’t have as good data for 100 years ago, but I suspect that mobility differences between the rich and poor were much greater then than they are today. Blue-collar relied mainly on foot travel, and farm families mainly on foot or horsepower, while wealthy and white-collar families had much greater access to mechanized travel such as intercity trains and streetcars. The point is that mobility today is much more evenly divided among income groups than it was before low-cost automobiles and airlines.

You can also make-your-own table showing trips by type of housing, including single-family, duplex, row house, and apartments/condos. Single-family households average 8.3 trips per day; duplexes 7.4 trips; rowhouses 5.4; and multifamily 6.0. So far so New Urbanist, but what happens when you account for household size? On a per-person basis, single-family households take 3.7 trips per day; duplexes 3.8; rowhouses 4.3; and multifamily 3.2. So rowhouses aren’t all that great.

Finally, I made a table showing trips per day by household for each of the 18 metropolitan areas identified in the data. The results offer little comfort to those who argue that increased densities will reduce travel. The highest of any metro area was Portland, at 9.2 trips per day. The lowest was Cincinnati (which urban area is fairly low density), at 7.4 trips per day. High-density Boston, Chicago, Los Angeles, and San Francisco-Oakland all scored higher than low-density Houston. Of course, in some of these cases the sample sizes were pretty small, but Houston, Los Angeles, and San Francisco-Oakland were all around 4,000 households or more.

So far I haven’t tried to download and analyze the full 500-megabyte data file. Excel could not handle this much data and I don’t have database software, though it might be worth getting some to review files like this. Any other insights into the data will be welcome in the comments.

Bookmark the permalink.

About The Antiplanner

The Antiplanner is a forester and economist with more than fifty years of experience critiquing government land-use and transportation plans.

5 Responses to 2009 National Household Travel Survey

  1. C. P. Zilliacus says:

    The Antiplanner wrote:

    > The first thing that must be noted is that the 150,000 households
    > surveyed were not an accurate cross-section of the nation.
    > For example, they surveyed one household for every 1,000 people
    > in the Dallas-Ft. Worth metropolitan area, but only one for
    > every 11,000 people in the Chicago metropolitan area,
    > and (apparently) no one at all in the Atlanta metro area.
    > Further, though more than two-thirds of Americans live in
    > urban areas of 50,000 people or more, only about 61 percent
    > of the surveys came from such areas.

    Hmmmm.

    They have to have included households in Atlanta! I cannot believe that a survey of this nature would exclude such a major metropolitan area.

    Now I do know that some of the apparent “uneven” sample sizes may be due to some states and some metropolitan planning organizations “buying” larger sample sizes in their states or regions.

  2. C. P. Zilliacus says:

    The Antiplanner also wrote:

    > So far I haven’t tried to download and analyze the full
    > 500-megabyte data file. Excel could not handle this much
    > data and I don’t have database software, though it might
    > be worth getting some to review files like this. Any other
    > insights into the data will be welcome in the comments.

    The right tool for a dataset of that size is SAS or perhaps SPSS.

    Both are industrial-strength data analysis and data mining tools (I personally prefer SAS because that’s what I know), but these packages are not cheap.

  3. Dan says:

    I agree with CPZ that either SAS or SPSS will do, depending upon what you want to slice and dice. Surely Cato or Fraser has ponied up for these packages and you can work remotely once you get back home (unless your laptop is your PC, in which case no problem). Contact IT and see how they can get you a connection to a PC with those programs.

    DS

  4. Scott says:

    The trips having no correlation w/density is no surprise.
    In the past, I have analyzed (Excel & SPSS) VMT, per capita & per driver, with about 6 different variables & have found only very slight correlations, which are certainly not consistent.

    My theory is that some of the main components of VMT are job distribution & housing distribution, with each having many variables within themselves, are different between persons, & cannot be measured quantitatively.

    The idea of neglecting roads does not help & has many unintended consequences: more congestion, slower econ growth, less pop growth, lower mpg, etc.

  5. dave says:

    These differences may explain some of the differences between the NHTS results and results from the American Community Survey, which the Census Bureau conducts of about 3 million households each year. For example, according to the 2008 ACS,2.9 percent of employees walk to work and 5.2 percent take transit to work. By comparison, the NHTS numbers indicate that 5.6 percent walk and only 3.7 percent take transit to work. This may reflect the NHTS bias towards less urban communities.

    This comparison of modal shares for walk and transit is inappropriate for the implied purpose. The ACS numbers are based on a worker’s usual mode usage (i.e., the usual mode to work in a week), but the NHTS numbers are based on a worker’s actual mode usage (i.e., individual work trips). The usual share is consistently smaller than the actual share for walk, but the reverse is true for transit. For more details, refer to the following:

    Chu and Polzin, Relationship Between Transit’s Usual and Actual Mode Shares, Transportation Research Record 2049, 14-20.

    Sikder and Chu, Is the Usual Share of a Commuting Mode always Greater than its Actual Share? TRB Paper No. 11-3388.

Leave a Reply